INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Iminosugar derivatives containing the glucose analogue deoxynojirimycin (DNJ) exert antiviral effects against viruses of different families, including Human immunodeficiency virus (HIV) (9-12, 16), Hepatitis B virus (HBV) (3, 18, 19), Woodchuck hepatitis virus (4), Bovine viral diarrhea virus (BVDV) (29), and Dengue virus (8). The antiviral effects were either proven (10, 11) or assumed to be associated with the inhibitory action of DNJ-containing iminosugar derivatives on enzymes in the endoplasmic reticulum (ER), -glucosidases I and II. DNJ, a ring-nitrogen-containing and unmetabolizable glucose analogue, competitively binds to these enzymes and prevents them from performing the stepwise removal of three glucose residues attached to the N-linked glycans carried by newly synthesized polypeptides. This in turn prevents these polypeptides from interacting with the ER chaperones calnexin and calreticulin, which bind to monoglucosylated glycoproteins (14). Interaction with these ER chaperones is crucial for the correct folding of some but not all glycoproteins (13, 17, 22). Potentially all viruses which encode glycoproteins that depend on calnexin interaction for proper folding could be targeted using ER -glucosidase inhibitors, with those budding from the ER (e.g., HBV and members of the Flaviviridae) probably being most sensitive (20).

In the absence of a reliable cell culture system able to support hepatitis C virus (HCV) replication and based on the degree of homology in terms of genomic organization, replication strategy (including chronicity), and protein function (25), BVDV has been proposed as a surrogate model to study the action of antiviral molecules (2, 29). We have previously shown that the BVDV envelope glycoproteins E1, E2, and the E2 precursor E2-p7 interact with calnexin and that this interaction can be inhibited using the iminosugar derivative N-butyl-DNJ (NB-DNJ) (5). The antiviral effect observed using the ER -glucosidase inhibitors NB-DNJ and N-nonyl-DNJ (NN-DNJ) was attributed to the reduction of viral secretion due to an impairment of viral morphogenesis caused by -glucosidase inhibition (29). However, a precise study of the mechanism of action at the molecular level has yet to be performed.

In this report we demonstrate that in addition to the antiviral effect caused by ER -glucosidase inhibition, iminosugar derivatives carrying longer alkyl side chains have a second mechanism of action. We further report the antiviral activity of long-alkyl-chain deoxygalactonojirimycins (DGJ), a class of iminosugars derived from galactose which lack ER -glucosidase inhibitory activity. Using both DNJ- and DGJ-containing iminosugar derivatives, we investigated their mechanism of action and eliminated the possibilities that they act at the level of replication, protein synthesis, or protein processing. On the basis of the effect of the length of the alkyl chain attached to the carbohydrate analogue head group on viral morphogenesis, secretion, and infectivity, we propose a second mechanism of action involved in the antiviral activity of long-alkyl-chain iminosugar derivatives. Finally, we show that the abnormal accumulation of E2-E2 dimers in the ER, induced by treatment with long-alkyl-chain iminosugars, and its subsequent enrichment in the virus particles correlate with reduced infectivity of these particles.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells, viruses, and inhibitors. Madin-Darby bovine kidney (MDBK) cells were maintained at 37°C in a humidified, 5% CO₂ atmosphere in RPMI 1640 medium (GIBCO/BRL) supplemented with 10% screened BVDV-free fetal calf serum (PAA Laboratories, Teddington, United Kingdom). The noncytopathic (ncp) Pe515 strain (23) and the cytopathic (cp) NADL (National Animal Disease Laboratory) strain were used in this study. The titers of the stock solutions of these viruses were determined to be 4 × 10⁶ and 2 × 10⁷ PFU/ml, respectively. NB-DNJ was a gift from Searle/Monsanto. NN-DGJ was purchased from Toronto Research Chemicals. 6-Deoxy-DGJ was kindly provided by G. W. Fleet (University of Oxford). NN-DNJ, N-nonyl-6-deoxy-DGJ, N7-oxadecyl-DNJ and N7-oxanonyl-6-deoxy-DGJ were either synthesized in-house or provided by Synergy Pharmaceuticals Inc.

Synthesis and purification of N-alkylated DNJ and DGJ compounds. N-alkylated-DNJ and -DGJ compounds were synthesized using sodium cyanoborohydride (NaBH₃CN) and the appropriate aldehyde (21, 29). The C₁₆ and C₁₈ compounds were prepared using cis-11-hexadecanal and cis-13-octadecanal, respectively, and purified as described previously (21).

MTS cell proliferation assay. MDBK cells were grown and incubated on 96-well plates at 37°C in the presence of different concentrations of inhibitors (in triplicate) for 3 days. The number of proliferating cells was determined using the CellTiter 96 AQ_ueous nonradioactive cell proliferation assay (Promega, Southampton, United Kingdom) as specified by the manufacturer. Cells were incubated with the 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS)-phenazine methosulfate (PMS) solution for 3 h at 37°C before the absorbance at 490 nm was read using an enzyme-linked immunosorbent assay plate reader. The 50% cytotoxic concentration (CC₅₀) was determined as the drug concentration at which half of the cells were not proliferating compared to untreated control cells.

BVDV plaque reduction assay. MDBK cells were grown in six-well plates to 90% confluency and infected with cp BVDV (at a multiplicity of infection [MOI] of 0.01, 0.1, or 1) for 1 h at 37°C. After removal of the inoculum, the cells were washed with phosphate-buffered saline (PBS) and incubated in RPMI 1640 medium containing 10% fetal calf serum and inhibitors at different concentrations. After 2 or 3 days, the medium containing secreted virus was removed from the wells and centrifuged at low speed to remove cellular debris. Different dilutions were used to infect a fresh monolayer of MDBK cells in six-well plates. After 2 days, the resulting plaques were counted. The 50 and 90% inhibitory concentrations (IC₅₀) or (IC₉₀) were determined as the concentrations at which the number of plaques was halved or reduced by 90%, respectively, compared to untreated infected control cells.

Analysis of viral genome synthesis. Subconfluent (90% confluent) MDBK cell monolayers grown in 25-cm² flasks were infected with BVDV at a MOI of >5. After 1 h, the inoculum was removed and the cells were washed twice in PBS before 5 ml of fresh medium containing inhibitors at the concentrations indicated was added. The inhibitors were present throughout the experiment. At 6 h postinfection (h p.i.), actinomycin D (5 µg/ml) was added to the medium to block host RNA synthesis and kept present for the rest of the experiment. At 8 h p.i., RNA was labeled by the addition of 100 µCi of [5,6-³H]uridine per ml to the medium. At 6 h after the onset of labeling, total RNA was extracted using the RNeasy purification kit (Qiagen, Crawley, United Kingdom) as specified by the manufacturer. RNA was fractionated by agarose-formaldehyde gel electrophoresis and transferred by capillary action onto Hybond N⁺ membrane (Amersham, Little Chalfont, United Kingdom) as described by Ausubel et al. (1). The membrane was treated with a fluorographic reagent (Amersham) for 30 min, dried, and exposed to Hyperfilm-MP (Amersham) at 70°C for 14 days. The intensity of the bands on the resulting autoradiogram was measured by scan densitometry.

Viral RNA purification and RT-PCR analysis. MDBK cells (8 × 10⁵/35-mm-diameter dish) were infected at an MOI of 1 with either cp or ncp BVDV. After 1 h, the inoculum was removed and the cells were washed twice in PBS before 2 ml of fresh medium containing inhibitors at the concentrations indicated was added. At 24 or 48 h p.i., the supernatant was harvested, centrifuged at 5,000 × g for 5 min, and stored at 70°C before being used for viral RNA purification. RNA from released viral particles was purified using the QIAamp viral RNA purification kit (Qiagen) with 140 µl of supernatant as the starting material. Reverse transcription-PCR (RT-PCR) was performed using the Titan One Tube RT-PCR system (Roche) and two primers, P1 (5'-AACAAACATGGTTGGTGCAACTGGT-3') and P2 (5'-CTTACACAGACATATTTGCCTAGGTTCCA-3'), which amplify a 826-bp region overlapping E^rns and E1 (27). A total of 35 cycles (10 cycles of 30 s at 94°C, 1 min at 55°C, and 1 min at 68°C and 25 cycles of 30 s at 94°C, 1 min at 55°C, and 1 min plus 5 s/cycle at 68°C) of PCR were performed. The PCR products were separated using 2% agarose gels stained with ethidium bromide.

Protein extraction, SDS-PAGE, and Western blot analysis. MDBK cells (8 × 10⁵/35-mm-diameter dish) were infected at an MOI of 1 with either cp or ncp BVDV. After 1 h, the inoculum was removed and the cells were washed twice in PBS before 2 ml of fresh medium containing inhibitors at the concentrations specified was added. At 24 or 48 h p.i., the cells were harvested by trypsin-EDTA treatment, washed once with PBS, centrifuged at 5,000 × g for 5 min, and stored as a dry pellet at 70°C before being used for protein extraction. Cell lysis was performed at 4°C for 1 h in CHAPS-HSE buffer (2% CHAPS, 50 mM HEPES [pH 7.5], 200 mM NaCl, 2 mM EDTA) in the presence of a cocktail of protease inhibitors (Sigma, St. Louis, Mo.) and 20 mM iodoacetamide (Sigma). Iodoacetamide alkylates free sulfhydryl groups and avoids nonspecific disulfide bond formation. The protein concentration of the cell lysates was determined using the bicinchoninic acid protein assay kit (Pierce, Rockford, Ill.). Protein samples were separated by sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) under nonreducing or reducing conditions and transferred onto nitrocellulose membranes. Three monoclonal antibodies (MAbs) directed against the BVDV E2 protein (MAbs WB214, WB166, and WB158 [Veterinary Laboratory Agency, Weybridge, United Kingdom]) were used as primary antibodies in this study. MAbs WB214 and WB166 recognize a linear epitope and are used under both nonreducing and reducing conditions. MAb WB158 is a conformation-dependent antibody, which recognizes E2 only under nonreducing conditions. Peroxidase-conjugated sheep anti-mouse immunoglobulin G (Amersham) was used as secondary antibody (diluted 1:2,000 in PBS). Proteins were detected using the ECL system as specified by the manufacturer (Amersham).

Pulse-chase labeling and immunoprecipitation. Subconfluent MDBK cell monolayers grown in 25-cm² flasks were infected with BVDV at an MOI of 1. After a 1-h incubation at 37°C, the viral inoculum was replaced with medium containing 10% FCS. At 18 h p.i., monolayers were washed once with PBS and incubated in methionine- and cysteine-free RPMI 1640 medium (ICN Flow). After 1 h, the cells were pulse-labeled with 100 µCi of [³⁵S]methionine-[³⁵S]cysteine (Tran ³⁵S-label, 1,100 Ci/mmol; ICN Flow) per ml at 37°C for the times indicated. Following labeling, the isotope-supplemented medium was removed and the cells were washed once with PBS and chased for various times in RPMI 1640 medium containing 10 mM unlabeled methionine. At the time points indicated, the chase media were discarded and the cells were harvested. When the effect of the iminosugar derivative on protein synthesis and protein processing was investigated, the drug was added to the cells at the concentrations indicated 2 h before the pulse and was present throughout the chase period. The cells were then lysed for 1 h on ice in a buffer containing 0.5% Triton X-100, 50 mM Tris-Cl (pH 7.5), 150 mM NaCl, 2 mM EDTA (Triton-TSE buffer), and a mixture of protease inhibitors. Labeled cell lysates were clarified by centrifugation at 12,000 × g for 15 min and precleared with 20 µl of protein G-Sepharose for 1 h at 4°C. The lysates were briefly centrifuged, and the supernatants were incubated with anti-BVDV E2 (MAb WB214; 1:50 dilution) overnight at 4°C. Protein G-Sepharose (30 µl) was then added to the supernatants, and the incubation was continued for 1 h at 4°C. The slurry was washed six times with 0.2% Triton X-100 in TSE buffer. The immunoprecipitated complexes were eluted by boiling the samples for 10 min in Laemmli buffer and separated by SDS-PAGE. After electrophoresis, the gels were treated with Amplifyer (Amersham), dried, and exposed at 70°C to Hyperfilm-MP (Amersham).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Antiviral features of newly developed iminosugar derivatives: importance of the head group and alkyl chain length. NB-DNJ and the longer-alkyl-side-chain-containing NN-DNJ have been reported previously to show an antiviral effect against BVDV in vitro (29). These experiments were performed using a low MOI (0.01). Since the MOI which is relevant with respect to an in vivo situation is unknown, we decided to investigate whether these molecules had antiviral properties at higher MOIs. We evaluated the antiviral effect of these two drugs and other iminosugar derivatives (Fig. 1) in BVDV plaque reduction assays using MDBK cells infected at different MOI (0.01, 0.1, and 1) (Table 1). The toxicity of the drugs determined in MTS assays was used to calculate the in vitro selectivity index (CC₅₀/IC₅₀) (Table 2). Introduction of an oxygen atom at position 7 of the alkyl side chain led to a major reduction in toxicity and significant improvement of the selectivity index.

View larger version (21K): [in this window] [in a new window]   FIG. 1.   Chemical structures of the iminosugar derivatives used in this study. (A) Iminosugar derivatives based on the glucose analogue DNJ. (B) Iminosugar derivatives based on the galactose analogue DGJ.

View larger version (27K): [in this window] [in a new window]   FIG. 2.   Antiviral effect of DNJ derivatives in the "in vitro" BVDV system. DNJ derivatives containing alkyl side chains of different lengths ranging from C4 to C18 (x axis) were tested for their ability to inhibit BVDV plaque formation on MDBK cells as described in Materials and Methods. An MOI of 0.01 was used. Hexadecan-DNJ and octadecan-DNJ are unsaturated. The IC50 of each molecule is shown on the y axis.

View larger version (32K): [in this window] [in a new window]   FIG. 3.   ER -glucosidase inhibition and antiviral effect of iminosugar derivatives. BVDV-infected MDBK cells (MOI of 1) were not treated (No) or were treated with NB-DNJ (NB) and NN-DNJ (NN) at their respective IC50s (left panel) and IC90s (right panel). Cells were lysed at 18 h p.i., and proteins were separated by SDS-PAGE (10% polyacrylamide) under reducing conditions. A Western blot analysis was performed using the anti-E2 MAb WB214 (diluted 1:1,000). The two polypeptides detected (E2 and E2-p7) are indicated by arrows.

Iminosugars do not inhibit viral RNA synthesis or the formation and processing of the viral polyprotein. Generally it has been assumed that DNJ derivatives exert their antiviral activity via -glucosidase inhibition only. We have shown that this is not the case for long-alkyl-chain DNJ derivatives and that an additional and as yet undefined mechanism is involved. It was therefore necessary to establish whether this second mechanism could target either BVDV RNA synthesis or the formation and processing of the BVDV polyprotein. The impact of the long-alkyl-chain-containing NN-DNJ and NN-DGJ, representing the two classes of iminosugar derivatives used in this study, on viral RNA synthesis was studied by monitoring the incorporation of [5,6-³H]uridine into newly synthesized BVDV genomic RNA in the presence or absence of the inhibitors. To this end, we used the method described previously by Purchio et al. (24) (see Materials and Methods). The amount of rRNA (28S and 18S) present in each lane was quantified and used to normalize the quantity of radio-labeled viral RNA. Neither NN-DNJ nor NN-DGJ treatment led to a significant reduction in the amount of radiolabeled genomic viral RNA, whereas control treatment with ribavirin, an inhibitor of viral replication, caused a significant reduction in the quantity of radiolabeled viral RNA detected (data not shown).

View larger version (19K): [in this window] [in a new window]   FIG. 4.   Biosynthesis and processing of BVDV envelope glycoproteins in the presence of NN-DNJ or NN-DGJ. MDBK cells were infected with BVDV at an MOI of 1. At 18 h p.i., the cells were not treated () or were treated (+) with 100 µM of NN-DNJ (A) or NN-DGJ (B). At 2 h later, the cells were pulse-labeled with [35S]methionine-[35S]cysteine for 15 min, chased for the times indicated in the continuous presence of the drug, and immunoprecipitated with anti-E2 MAb WB214. The proteins were separated by SDS-PAGE (10% polyacrylamide) under reducing conditions and visualized by autoradiography. The two main polypeptides detected (E2 and E2-p7) are indicated by arrows.

DNJ but not DGJ derivatives impair the secretion of virus particles. Previously we assumed that all of the antiviral effect observed with NB-DNJ and NN-DNJ was due to the inhibition of secretion of virions into the culture medium (29). The experiments leading to this conclusion had been performed at an MOI of 0.01 in a 3-day plaque reduction assay, i.e., under conditions which were not optimal to study the secretion of viral RNA. We therefore analyzed the effects of both DNJ and DGJ derivatives on the secretion of virions by using MDBK cells infected at a higher MOI. Cells infected with the cytopathic NADL strain of BVDV at an MOI of 1 were treated with both DNJ and DGJ derivatives. At 24 h p.i., the level of enveloped viral RNA released was measured by RT-PCR. The viral RNA was purified either directly from culture medium supernatants or from the virus-enriched fraction pelleted through a 20% sucrose cushion. To achieve accurate quantification by the end-point RT-PCR method, fivefold dilutions of the RNA samples were subjected to the same number of PCR cycles (15, 27). At the same time, an RT-PCR standard curve was obtained using supernatants obtained by twofold serial dilutions of a virus stock with a known titer of 2 × 10⁷ PFU/ml. The RT-PCR products were separated by agarose gel electrophoresis (2% agarose) and visualized by ethidium bromide staining (Fig. 5A). Densitometry analysis of the gel revealed a 5.5- and 4.5-fold reduction of the signal (compared to the untreated control) after NB-DNJ (2.5 mM) and NN-DNJ (100 µM) treatment, respectively. No reduction was observed after treatment with NN-DGJ. Identical results were obtained using RNA purified directly from the supernatant or purified from material pelleted through a 20% sucrose cushion, indicating that unenveloped RNA, which could have led to a misinterpretation of the result, was not present. These results indicate that DNJ but not DGJ derivatives impair the secretion of virus particles.

View larger version (67K): [in this window] [in a new window]   FIG. 5.   Effect of iminosugar derivatives on viral secretion. MDBK cells were infected with either a cp (NADL) (A) or ncp (Pe515) (B) strain of BVDV at an MOI of 1 and grown for 24 h (A) or 48 h (B) in the absence or presence of different iminosugar derivatives at the concentrations indicated. Viral RNA was purified from the supernatant and used to perform a single-tube RT-PCR analysis. Then 10 µl (A) or 20 µl (B) of the RT-PCR product was loaded and run on a 2% agarose gel stained with ethidium bromide. (A) Result obtained using the cp NADL strain. The quantity of cDNA was determined by densitometry analysis. The experimental titer was measured by the plaque reduction assay, while the theoretical titer was calculated using the linear regression curve obtained with the standard curve (established from a virus stock with a titer of at 2 × 107 PFU/ml). A summary of these results is presented in the table shown below the gels. (B) Result obtained using the ncp Pe515 strain. The quantity of cDNA was determined by densitometry analysis, and the experimental titer was measured by the plaque reduction assay. The values corresponding to the reduction in cDNA quantity and the reduction in the titer, obtained by comparison of treated and nontreated samples, are presented in the table below the gels.

Treatment with long-alkyl-chain DNJ derivatives and DGJ derivatives leads to the creation of viral particles with reduced infectivity, which contributes to the antiviral effect observed. The standard curve in the experiment described above was used to determine the theoretical titer (by comparing the cDNA quantities of treated and untreated samples with those of serial dilutions of a virus stock of known titer) for each of the samples analyzed. This was then compared to the actual titer established in plaque reduction assays. The experimentally determined titers were 7 × 10⁵ PFU/ml for nontreated, 6 × 10⁴ PFU/ml for NB-DNJ treated, 8 × 10⁴ PFU/ml for NN-DNJ-treated, and 1 × 10⁵ PFU/ml for NN-DGJ-treated sample supernatants (table in Fig. 5A). This represented a 11.65-, 8.75-, and 7.1-fold drop in the titer for NB-DNJ-, NN-DNJ-, and NN-DGJ-treated samples, respectively. Compared to the theoretical titer (table in Fig. 5A), these numbers indicated a discrepancy between the quantity of viral RNA detected and the actual titer of infective virus. The results indicate that DNJ derivatives inhibit the secretion of viral particles to a certain extent. This is, however, not sufficient to account for all of the antiviral effect observed. However, the particles that do get secreted are not as infective as wild-type particles, which accounts for the stronger antiviral effect observed. Long-alkyl-chain DGJ derivatives do not inhibit the secretion of viral RNA. Their antiviral effect is entirely due to a second mechanism, which reduces the infectivity of the particles released.

Iminosugar derivatives show the same antiviral effect against the cp and the ncp biotypes of BVDV. BVDV strains fall into two biotypes which differ according to their pathogenicity in cultured cells. The ncp strains cause no visible in vitro pathogenicity, whilst the cp strains cause cell death by apoptosis (28). The cp virus was used for antiviral screening, since it induces plaques in the host cell monolayers, which can be easily quantified. However, the virological features of the ncp biotype may more closely resemble those of HCV, the ultimate proposed target for the drugs developed. Therefore, we repeated the RT-PCR experiment described above using the ncp Pe515 strain. Ribavirin, which inhibits viral RNA replication, was used as a control. Treatment with ribavirin led to a 30-fold decrease in the amount of viral RNA secreted from cells, which correlates with the 32-fold reduction in infectious viral titer (Fig. 5B). Mirroring the results obtained with the cp biotype, there was a discrepancy between the theoretical and experimentally determined viral titer of samples treated with DNJ and DGJ derivatives. The two DNJ derivatives tested, NN-DNJ and N7-oxadecyl-DNJ (chemical structures in Fig. 1), inhibited the secretion of viral particles, as measured by the reduction of viral RNA levels in the supernatants (table in Fig. 5B) 5- and 3-fold, respectively. This alone could not explain the 13.5- and 6.35-fold reduction in the amount of infectious virus. Treatment with NN-DGJ did not lead to a significant reduction (1.2-fold drop) of secreted viral RNA, but it did lead to a 7.5-fold reduction in infectious viral titer. The results obtained using the ncp biotype confirmed those achieved with the cp biotype. They indicate that part of the antiviral effect achieved with long-alkyl-chain DNJ derivatives and all of the antiviral effect achieved with long-alkyl-chain DGJ derivatives is due to the creation of virus particles with reduced infectivity.

Long-alkyl-chain DNJ derivatives inhibit viral RNA secretion in a dose-dependent manner. The RT-PCR experiments described above were performed at fixed inhibitor concentrations. To observe the dose-response effect, we chose iminosugar derivatives which combined good antiviral efficacy with low toxicity. N7-Oxadecyl-DNJ and N7-oxanonyl-6deoxy-DGJ, representing the two classes of iminosugar derivatives used in this study, have IC_{50[MOI 1]}s of ~80 and 125 µM, respectively. The oxygen atom in the side chain results in a comparatively low in vitro toxicity, with CC₅₀s of ~5 and 4 mM, respectively. The reduced toxicity allowed the use of higher inhibitor concentrations in the assay and the establishment of a dose-response curve. The experiment was performed using both the cp and ncp strain of BVDV. The virus-derived RT-PCR signal decreased with increasing concentrations of N7-oxadecyl-DNJ (Fig. 6), indicating an increasing inhibition of viral RNA secretion. With N7-oxanonyl-6deoxy-DGJ, only a slight reduction of viral RNA secretion was observed at the highest inhibitor concentration used (5 mM). However, since the CC₅₀ of this compound lies between 4 and 5 mM, this reduction is most probably caused by general toxicity of the compound and not by a specific antiviral mechanism. These results show that long-alkyl-chain DNJ derivatives inhibit viral RNA secretion in a dose-dependent manner. They further show that while N7-oxanonyl-6deoxy-DGJ reduces the infectious viral titer, it does not reduce the amount of viral RNA secreted. This confirms the data obtained with NN-DGJ, showing that long-alkyl-chain DGJ derivatives exert their antiviral effect solely via the production of virus particles with reduced infectivity.

View larger version (74K): [in this window] [in a new window]   FIG. 6.   Effect of iminosugar derivatives on viral secretion: a dose-response analysis. MDBK cells were infected with either a cp (NADL) or ncp (Pe515) strain of BVDV at an MOI of 1 and grown for 24 h (cp) or 48 h (ncp) in the absence or presence of increasing amounts of two different iminosugar derivatives. Viral RNA was purified from the supernatant and used to perform a single-tube RT-PCR analysis, as described in Materials and Methods. Then 10 µl (cp) or 20 µl (ncp) was loaded and run on an ethidium bromide-stained 2% agarose gel. Computerized-inverted images of the gels are presented. The type of virus and drug used are indicated in boxes within the frames of the gels.

DNJ and DGJ derivatives with long alkyl side chains change the pattern of envelope glycoprotein dimer formation in the ER of infected cells. Our investigations into the cause of the formation of less infectious particles focused on the viral envelope glycoproteins and their dimerization in the ER. We have shown previously that treatment with NB-DNJ causes misfolding of the envelope proteins (5). The antiviral effect of NB-DNJ correlated with the misfolding of BVDV envelope glycoproteins and the impairment of their association into E1-E2 heterodimers. For the shorter-alkyl-side-chain derivative NB-DNJ, this effect was entirely due to -glucosidase inhibition mediated by the iminosugar headgroup, since the galactose analogue with the same alkyl chain length (NB-DGJ) did not show any antiviral effect. In this study we investigated the extent to which the longer-alkyl-chain DNJ derivatives impair glycoprotein dimer formation in the ER. DGJ derivatives were used as a control. MDBK cells infected with ncp BVDV (MOI of 1) were treated with 100 µM NN-DNJ, N7-oxadecyl-DNJ, NN-DGJ, or N7-oxanonyl-6deoxy-DGJ. At 48 h p.i., whole-cell lysates were prepared and analyzed under nonreducing conditions by SDS-PAGE followed by Western blotting and probing with MAb WB166, which recognizes a linear E2 epitope.

View larger version (49K): [in this window] [in a new window]   FIG. 7.   Effect of iminosugar derivatives on the accumulation of E2-E2 homodimers and E1-E2 heterodimers in the ER. (A) MDBK cells were mock infected (lane 1) or infected with the ncp strain (Pe515) at an MOI of 1 and grown in the absence (lane 2) or presence of 100 µM iminosugar derivatives (NN-DNJ, lane 3; N7-oxadecyl-DNJ, lane 4; NN-DGJ, lane 5; N7-oxanonyl-6deoxy-DGJ, lane 6). Ribavirin, which inhibits viral RNA replication, was used as a control (lane 7). At 48 h p.i., the cells were lysed and proteins were separated by SDS-PAGE (10% polyacrylamide) under nonreducing conditions. A Western blot analysis was performed using MAb WB166 (diluted 1:1,000), which recognizes the E2 glycoprotein regardless of its state of folding. In the bottom panel, the same blot was hybridized with an anti-actin antibody (loading control). (B) The same experiment as described for panel A was performed, but this time both cp and a ncp strains of BVDV were used for the infection. N7-oxadecyl-DNJ (abbreviated N7-DNJ) and N7-oxanonyl-6deoxy-DGJ (abbreviated N7-DGJ) were used at the concentrations indicated. The autoradiographs were quantified. The ratio of the amount of E2-E2 to the amount of E1-E2 was established, and the percentages, with the ratio of the amount of E2-E2 to the amount of E1-E2 without inhibitor present set at 100%, are shown at the bottom of each gel. Size standards are indicated on the left, and immunoreactive proteins and complexes are indicated on the right.

View larger version (27K): [in this window] [in a new window]   FIG. 8.   Correlation between the presence of a long alkyl chain branched on DNJ or DGJ head groups and the abnormal accumulation of E2-E2 dimers. MDBK cells were mock infected or infected with the cp (NADL) strain of BVDV at an MOI of 1 and grown for 24 h in the absence or presence of iminosugar derivatives at the concentrations indicated. The cells were lysed, and proteins were separated by SDS-PAGE (10% polyacrylamide) under nonreducing conditions. A Western blot analysis was performed using MAb WB166 (diluted 1:1000). The IC50 and IC90 of each drug are indicated by arrows. No IC50 or IC90 is indicated for NB-DGJ, since this compound is not antiviral.

The envelope glycoprotein dimer composition of the virion changes on treatment with long-alkyl-chain iminosugar derivatives. To determine whether the change in the ratio of E1-E2 to E2-E2 dimers synthesized within drug-treated cells was reflected in released virions, a Western blot analysis was performed with virus particles harvested directly from the cell culture medium. A total of 4 × 10⁷ MDBK cells were infected at an MOI of 1 and treated with different drugs (NB-DNJ at 2.5 mM and NN-DNJ and NN-DGJ at 150 µM) or left untreated. After 24 h, the medium (90 ml in total) was harvested, centrifuged at 5,000 × g for 30 min to separate viral particles from cells, split in two, and processed by polyethylene glycol (PEG) 8000 precipitation or by centrifugation through a 20% sucrose cushion. Proteins from pellets were separated by SDS-PAGE under nonreducing conditions and subjected to Western blot analysis using an anti-E2 antibody (MAb WB166) (Fig. 9). When virus-infected cells were treated with NB-DNJ or NN-DNJ, the quantity of total virion-associated E2 released from cells was reduced, confirming the inhibition of secretion shown in Fig. 5A. This reduction was more pronounced with the short-alkyl-side-chain-containing NB-DNJ than with NN-DNJ, confirming the trend observed in the RT-PCR assay (Fig. 5). After treatment with NN-DGJ, a slight increase in the quantity of total E2 was detected. This correlated with the increased accumulation of E2 observed in cells and with the fact that NN-DGJ did not inhibit viral secretion. The relative quantity of E1-E2 and E2-E2 was determined by computer-assisted densitometry analysis. The percentage reflecting the increase of the quantity of E2-E2 with respect to the quantity of E1-E2 was calculated (table in Fig. 9). The results indicated that compared to untreated infected cells, proportionally more E2-E2 homodimers were present in virus particles released from drug-treated cells, suggesting that drug treatment induced a change of the envelope glycoprotein dimer composition of the virus particles. For NN-DGJ, which, unlike NB-DNJ and NN-DNJ, does not impair the secretion of viral particles (Fig. 5), the change in the viral envelope composition directly correlated with the creation of viral particles with reduced infectivity and could explain the drop in infectious titer. For NB-DNJ and NN-DNJ, the antiviral effect observed was due both to an inhibition of the secretion of virus particles and the production of particles with reduced infectivity. The latter correlated with the change in the glycoprotein dimer composition of the viral envelope.

View larger version (58K): [in this window] [in a new window]   FIG. 9.   Analysis of the envelope glycoprotein dimer composition of the virus particles released under drug treatment. MDBK cells were infected with the cp (NADL) strain of BVDV at an MOI of 1 and grown in the absence (lanes 1 and 1') or presence of NB-DNJ (2.5 mM; lanes 2 and 2'), NN-DNJ (150 µM; lanes 3 and 3'), or NN-DGJ (150 µM; lanes 4 and 4'). After 24 h, the medium was harvested, clarified at 5,000 × g for 30 min, split in two, and processed by PEG 8000 precipitation (10,000 × g) or by centrifugation (90,000 × g) through a 20% sucrose cushion. Proteins from pellets were separated by SDS-PAGE (10% polyacrylamide) under nonreducing conditions and subjected to a Western blot analysis using the anti-E2 MAb WB166. The bands detected were quantified by densitometry. The ratio of the amount of E1-E2 to the amount of E2-E2 was established for every drug concentration used, and the percentages, with the ratio of the amount of E2-E2 to the amount of E1-E2 without inhibitor present set at 100%, are shown in the table below the gel. a, values are combined quantities (determined by densitometry analysis) from both gels; b, values were taken from the previous column.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The use of iminosugars containing the glucose analogue DNJ as antiviral agents against different viruses has been suggested since the late 1980s (3, 4, 8-12, 16, 18, 19, 29). While the action of two of them, DNJ and NB-DNJ, has been extensively described in the literature, the discovery of the antiviral action of a longer-alkyl-chain derivative of DNJ, NN-DNJ, was reported only recently (29). In the first part of this report we presented antiviral and toxicity data for newly developed iminosugar derivatives. Several major discoveries were made during this phase of development and research, which was aimed at finding more potent and less toxic antivirals. While investigating whether the increased antiviral activity observed with NN-DNJ (compared to that of NB-DNJ) was due to an effect mediated by its long alkyl side chain, the C₉ derivative of DGJ, NN-DGJ, was synthesized and found to be as effective as NN-DNJ in antiviral tests. This raised an interesting question about the mechanism of action of this molecule and also led us to doubt that the antiviral effect observed with NN-DNJ was indeed entirely due to the inhibition of the ER -glucosidases, as assumed previously. Because the shorter-alkyl-side-chain derivative NB-DGJ did not have any antiviral activity, it was hypothesized that the length of the alkyl side chain played an important role. However, nonylamine on its own did not show any specific antiviral activity, indicating that both the DGJ head group and the C₉ alkylchain were responsible for the antiviral effect observed.

Efficacy and toxicity are the two most important aspects in the search for potential antivirals, with toxicity often being the limiting factor. Judged by their selectivity index (CC₅₀/IC₅₀), the best iminosugar derivative of each class tested in the in vitro BVDV system were N7-oxadecyl-DNJ and N7-oxanonyl-6deoxy-DGJ. The latter has the added advantage of being only a very weak inhibitor of the ceramide-specific glucosyltransferase (Butters, unpublished), which would minimize side effects related to the inhibition of glycosphingolipid synthesis. Furthermore, the lowered toxicity, which is due to the oxygen atom at position 7 in the alkyl chain of both N7-oxadecyl-DNJ and N7-oxanonyl-6deoxy-DGJ, may lead to improved in vivo results.

While the antiviral mechanism of NB-DNJ has been extensively studied using both BVDV (5) and HIV (9-11) as models, the molecular details of the antiviral action of long-alkyl-chain derivatives of DNJ, such as NN-DNJ, and of the newly identified long-alkyl-chain derivatives of DGJ were unknown. Using the method described by Purchio et al. (24) to measure the synthesis of viral genomic RNA and using pulse-chase experiments to analyze protein synthesis and processing, we have demonstrated that long-alkyl-chain iminosugar derivatives do not interfere with either the BVDV RNA replication process or viral protein synthesis and processing. Our previous work suggested that DNJ derivatives inhibit the secretion of BVDV (29). After optimizing the experimental conditions, we revisited this hypothesis and measured the effect of iminosugar derivatives on viral secretion during a single round of high-MOI infection by quantitative RT-PCR. Our results show that DNJ-based derivatives reduce the secretion of RNA-containing virus particles, while no decrease of the level of enveloped viral RNA occurs with DGJ-based derivatives, emphasizing a fundamental difference in the mechanism of action of these two classes of molecules. Treatment with NB-DNJ impairs the proper folding of E1 and E2 glycoproteins and their association into heterodimers, which is a direct effect of the ER -glucosidase inhibition (5). Although to a lesser extent, the longer-alkyl-side-chain derivative NN-DNJ also inhibits the ER -glucosidases (Fig. 3, 4, and 7), and the impaired viral secretion observed with this inhibitor is likely to be caused by the same effect of impaired heterodimer formation as seen with NB-DNJ. However, after treatment with either NB-DNJ or NN-DNJ, a discrepancy became apparent between the reduction in the quantity of secreted viral RNA and the reduction in titer. Such a discrepancy was not observed when ribavirin, an inhibitor acting at the level of RNA replication, was used in control experiments. These results suggest that the inhibition of secretion observed with NB-DNJ and NN-DNJ does not account for the entire antiviral effect caused by these molecules. Here we provided evidence that the creation of viral particles with reduced infectivity contributes to the antiviral effect to different extents, depending on the alkyl chain length attached to the iminosugar headgroup. For longer-alkyl-chain-DGJ-based derivatives, the entire antiviral effect is mediated via the production of less infectious viral particles. Most of the antiviral effect observed with NN-DNJ and even some of the antiviral effect observed with the shorter-alkyl-chain derivative NB-DNJ is also due to the creation of particles of reduced infectivity. A first clue to what may be the cause of this reduced infectivity came when we analyzed the glycoprotein dimer composition of secreted viral particles. We found that in comparison to virions released from untreated cells, the ratio between E1-E2 heterodimers and E2-E2 homodimers had changed. These results were confirmed by observations made with material derived from infected cells. The long-alkyl-chain compounds, regardless of whether they carry a DGJ or DNJ head group, induce an accumulation of E2-E2 homodimers (by an unknown mechanism). The short-alkyl-chain compound NB-DNJ does not induce such an accumulation. However, as an ER -glucosidase inhibitor, NB-DNJ causes a decrease in the E1-E2 heterodimer formation and, by reducing the number of E1-E2 dimers, also changes the ratio between E2-E2 and E1-E2 dimers. The long-alkyl-chain compound NN-DNJ is also an ER -glucosidase inhibitor and combines both effects. The E2-E2 accumulation caused by the presence of the long alkyl chain may contribute predominantly to its antiviral effect. However, the additional antiviral effect caused by the ER -glucosidase inhibition may be the reason why DNJ-based compounds consistently achieve better IC₉₀s. Whether the changed glycoprotein dimer composition in secreted viral particles only reflects or indeed accounts for the reduced infectivity of these particles will have to be established. In the absence of proper tools to investigate the behavior of the other two envelope glycoproteins, E1 and E^rns, the overall picture of the antiviral mechanism of action is still far from complete.

In cells treated with long-alkyl-chain iminosugar derivatives (NN-DNJ and NN-DGJ), the overall quantity of E2 protein increases, which is most noticeable in E2-E2 homodimers. Since we have shown that neither viral RNA replication nor protein synthesis is modified by these iminosugars, a differential protein degradation rate may be responsible for this phenomenon. In this model, treatment with long-alkyl-chain iminosugar derivatives would lead to a delayed viral protein degradation. However, viral protein degradation may not be affected in cells treated with shorter-alkyl-chain iminosugar derivatives, which would explain the absence of overall accumulation of E2 protein in NB-DNJ treated cells. This hypothesis remains to be tested.